Re: Boot code broken in some way in recent current (Was: Panic at boot)

From: Roman Kurakin <rik_at_cronyx.ru>
Date: Tue, 24 Aug 2004 00:15:52 +0400
John Baldwin:

>On Saturday 21 August 2004 04:35 pm, Doug White wrote:
>  
>
>>On Fri, 20 Aug 2004, Roman Kurakin wrote:
>>    
>>
>>>    At first I have two SMP system (Dual) one that boots normaly and one
>>>that
>>>goes to panic (the older pc).
>>>   I've setup finaly serial console to that system, so here is more
>>>detailed output:
>>>
>>>OK boot -Dhv
>>>/boot/kernel/acpi.ko text=0x406a8 data=0x1bc4+0x110c
>>>syms=[0x4+0x72d0+0x4+0x9754]
>>>      
>>>
>>Have you tried booting without enabling ACPI?  The ACPI tables seem to
>>think you have two APs and no BSP.
>>    
>>
>
>ACPI tables have no concept at all of a BSP.  We infer the BSP by checking the 
>APIC ID of the CPU we are currently executing on when enumerating CPUs.
>  
>
>>Also check the BIOS and set the MPTable version to 1.4, and upgrade the
>>BIOS itself if there is an update.
>>    
>>
>
>Since it worked before the recent loader change, I doubt this will make a 
>difference.  More likely is that the code picked a bad memory address to 
>install the AP trampoline into for some reason.
>
Bad from which point of view? From hardware or software?
How could I check you idea?

This is main machine I use for testing of mpsafe locking for sppp(4), cx(4),
ctau(4), cp(4). Since this is a singles SMP machine that has ISA bus that I
have. And I really hope to get these changes to 5.3.

Fixing this bug is a part of my daily job for now. I'll try to spend 
some time
to find out reason of this problem by my self (hour or two every day). But
since this part of code was out of my interest before it would be very
difficult.

So if you have an idea what to check and were to look based on your
knowledge of code I'll be glad to move this way.

rik
Received on Mon Aug 23 2004 - 18:29:08 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:08 UTC